Prior packet transmission and switching systems and/or networks included fault recovery arrangements. One such prior arrangement required a so-called centralized network management function to analyze reported faults and to reconfigure the network as required. Consequently, the centralized network management function required knowledge of the entire network and connectivity to each node in the network. Such arrangements are slow to respond to faults and are also susceptible to faults in the network and in the management function itself.
In another prior fault recovery arrangement, each node in the network required knowledge of the network configuration and knowledge of faults occurring in the network. In such an arrangement, each node must store additional network configuration data other than that needed for transmission of packets in the particular node. Any change in the network configuration may require a change in the formation being stored in the node. Both the storing and updating of the configuration information in the node is cumbersome and expensive to implement.
More recently, recovery from faults in a packet network has been realized by detecting faults in the transmission path associated with a network node and transmitting a fault indication message blindly for each virtual link that is affected by the fault. See, for example, the co-pending U.S. Pat. application Ser. No. 07/431,797 which was filed concurrently herewith.
Although these arrangements function satisfactorily in many applications, they may not function satisfactorily in others. Indeed, a problem with such arrangements occurs in complex networks that contain a large diversity of paths and/or uni-directional routing. For example, the previous arrangements may not properly respond to multiple faults in the network since the entire virtual circuit affected by the fault is switched to an alternate path even though only a portion or portions of the circuit are actually faulted. Furthermore, if the fault affects a uni-directional virtual circuit, the fault indication messages in the previous arrangements may not reach the originating node.